Parallel computing method of deep belief networks and its application to traffic flow prediction
作者：
Highlights：
• A parallel computing method for the DBN pre-training and fine-tuning phases is proposed.
• The parallel computing method is based on the master–slave and data parallel processing structure.
• The parallel computing method is applied to traffic flow prediction.
• Experimental results testify the effectiveness of the parallel computing method.
• The performances are compared between the serial and parallel computing methods.
摘要
•A parallel computing method for the DBN pre-training and fine-tuning phases is proposed.•The parallel computing method is based on the master–slave and data parallel processing structure.•The parallel computing method is applied to traffic flow prediction.•Experimental results testify the effectiveness of the parallel computing method.•The performances are compared between the serial and parallel computing methods.
论文关键词：Deep learning,Deep belief network,Parallel computing,Traffic flow prediction
论文评审过程：Received 29 July 2018, Revised 12 October 2018, Accepted 14 October 2018, Available online 19 October 2018, Version of Record 21 November 2018.
论文官网地址：https://doi.org/10.1016/j.knosys.2018.10.025
